Analysis of promoter regions in the saxitoxin gene cluster identifies potential bottlenecks in heterologous biosynthesis

Background: Dolichospermum circinale is a filamentous bloom-forming cyanobacterium responsible for biosynthesis of the paralytic shellfish toxins (PST), including saxitoxin. PSTs are neurotoxins and in their purified form are important analytical standards for monitoring the quality of water and seafood and biomedical research tools for studying neuronal sodium channels. More recently, PSTs have been recognised for their utility as local anaesthetics. Characterisation of the transcriptional elements within the saxitoxin ( sxt ) biosynthetic gene cluster (BGC) is a first step towards accessing these molecules for biotechnology. Results: In D. circinale AWQC131C the sxt BGC is transcribed from two bidirectional promoter regions encoding five individual promoters. These promoters were identified experimentally using 5ʹ RACE and their activity assessed via coupling to a lux reporter system in E. coli and Synechocystis sp. PCC 6803. Transcription of the predicted drug/metabolite transporter (DMT) encoded by sxtPER was found to initiate from two promoters, P sxtPER and P sxtPER2 . In E. coli , strong expression of lux from P sxtP , P sxtD and P sxtPER was observed while expression from P orf24 and P sxtPER2 was remarkably weaker. In contrast, heterologous expression in Synechocystis sp. PCC 6803 showed that expression of lux from P sxtP , P sxtPER , and P orf24 promoters was statistically higher compared to the non-promoter control, while P sxtD showed poor activity under the described conditions. Conclusions: Both of the heterologous hosts investigated in this study exhibited high expression levels from three of the five sxt promoters. These results indicate that future experiments to clone the complete sxt BGC into either heterologous host should result in the transcription of the complete pathway following engineering of the least active promoters. Therefore, heterologous expression of the sxt BGC in either E. coli or Synechocystis could be a viable option for producing PSTs for industrial or biomedical purposes. per OD Synechocystis PCC6803. One-way ANOVA (Graphpad statistical differences between the promoters. Unpaired t-tests were also used to determine the statistical differences between the strains and the control.

6803. Transcription of the predicted drug/metabolite transporter (DMT) encoded by sxtPER was found to initiate from two promoters, P sxtPER and P sxtPER2 . In E. coli , strong expression of lux from P sxtP , P sxtD and P sxtPER was observed while expression from P orf24 and P sxtPER2 was remarkably weaker. In contrast, heterologous expression in Synechocystis sp. PCC 6803 showed that expression of lux from P sxtP , P sxtPER , and P orf24 promoters was statistically higher compared to the nonpromoter control, while P sxtD showed poor activity under the described conditions.
Conclusions: Both of the heterologous hosts investigated in this study exhibited high expression levels from three of the five sxt promoters. These results indicate that future experiments to clone the complete sxt BGC into either heterologous host should result in the transcription of the complete pathway following engineering of the least active promoters. Therefore, heterologous expression of the sxt BGC in either E. coli or Synechocystis could be a viable option for producing PSTs for industrial or biomedical purposes. Background Saxitoxin (STX) is a neurotoxin produced by cyanobacteria and microalgae, a member of the broader group of alkaloids known as the paralytic shellfish toxins (PSTs) (Wiese et al., 2010). When high concentrations of PSTs are consumed by humans, acute poisoning can lead to death due to respiratory paralysis (Etheridge, 2010). Therefore, PSTs are needed as analytical standards for the monitoring and protection of commercial seafood and freshwater reservoirs, as well as for use in biomedical research. While the PSTs pose a significant public health risk and economic burden on society during algal bloom events, their scientific and pharmaceutical potential is well known (Adams et al., 1976, Lagos, 2014. Purified PSTs have been a critical tool for researchers investigating neuronal sodium channels, where the toxins specifically block site 1 of voltage-gated sodium channels (Cestèle & Catterall, 2000). Under controlled administration, PSTs are potent anaesthetics, particularly in combination with other local anaesthetics (Barnet et al., 2004, Rodriguez-Navarro et al., 2007. Further attempts to utilise STX in clinical trials are hindered by its toxicity, but more recent approaches, such as generating liposomal formulations of STX, resulted in blockage of the sciatic nerves in rats with no myotoxic, cytotoxic or neurotoxic effects (Epstein-Barash et al., 2009). It has been postulated that the same delivery will provide effective treatment for local and severe joint pain (Chorny & Levy, 2009). Other PSTs, such as the gonyautoxins, also have clinical potential and have been utilised for the treatment of anal fissures and chronic tension type headache (Garrido et al., 2004, Garrido et al., 2005, Lattes et al., 2009. Obtaining significant quantities of purified saxitoxins for clinical research or water quality analysis is difficult. Additionally, the chemical synthesis of PSTs is complex and unfeasible due to low yields, hence the most common form of obtaining purified compounds involves extraction and isolation from dinoflagellate blooms, cyanobacterial cultures or contaminated shellfish (Laycock et al., 1994).
Although analytical calibration standards are commercially available from the National Research Council Canada (NRC), shortages have hampered implementation of routine monitoring of PSTs in bivalves, forcing a reliance on mouse bioassays. These issues clearly highlight the necessity for a feasible and reliable method of commercial PST production and purification.
E. coli is a suitable host for heterologous expression of cyanobacterial pathways based on its fast growth rate and similar GC content. Initial studies used native promoters to produce the ribosomal peptides patellamide A and C, and the microviridins (Long et al., 2005, Ziemert et al., 2008. Recently, there has been a focus on the heterologous expression of cyanobacterial natural product 4 pathways in E. coli using the tetracycline-inducible Ptet O promoter (Liu et al., 2017, Greunke et al., 2018. The lyngbyatoxin biosynthetic gene cluster (ltx) has also been expressed in the cyanobacterium Anabaena sp. PCC7120 and E. coli GB05-MtaA (Ongley et al., 2013, Videau et al., 2016. While the native ltx promoters were successfully activated in Anabaena sp. PCC7120 to produce lynbyatoxin A, the native promoters were not activated in E. coli. Addition of Anabaena sp. PCC7120 sigma factors to the E. coli host also failed to induce expression of lyngbyatoxin A, suggesting that the heterologous host was unable to recognise the cyanobacterial ribosome binding sites (Wells et al., 2018). While growth of cyanobacteria is much slower than other common microorganisms, they remain a valuable source of significant natural products, and therefore, greater understanding of cyanobacterial promoters is essential. The disconnect between promoter efficiency in different organisms remains poorly understood and therefore, it is essential that promoters are compared in the host organism and heterologous vector (Huang et al., 2010) for biotechnological applications.
For the most part, cyanobacterial transcription machinery conforms to the system in E. coli, with the main difference being the widespread absence of the -35 hexamer in cyanobacteria, replaced by a transcription factor binding site to initiate transcription (Curtis & Martin, 1994). In E. coli, σ 70 is able to recognise the majority of promoters while in cyanobacteria, a range of sigma factors have been identified. (Imamura et al., 2003, Imamura & Asayama, 2009).
The saxitoxin biosynthetic (sxt) gene cluster has recently been characterised in six cyanobacterial species from the order Nostocales and one from the order Oscillatoriales (Kellmann et al., 2008, Mihali et al., 2009, Stucken et al., 2010, Mihali et al., 2011, Cullen et al., 2018. Each sxt cluster encodes a 'core' set of enzymes putatively responsible for STX biosynthesis, supplemented with 'tailoring' and 'auxiliary' genes that give rise to PST analogues or perform functions after PST biosynthesis.
Here, we characterise the transcription of sxt within the cyanobacterium Dolichospermum. circinale AWQC131C. Transcriptional units were identified which enabled the experimental isolation of two promoter regions. Activity of the promoters was assayed in the heterologous hosts E. coli and Synechocystis sp. PCC6803 using a lux reporter. Characterisation of these cyanobacterial promoters and determination of their activity in E. coli and Synechocystis provides novel expression strategies for PST biosynthesis with biotechnological, pharmaceutical and analytical applications.

Results And Discussion
Identification of transcription units within the sxt gene cluster Reverse-transcriptase PCR revealed that the encoded enzymes of PST biosynthesis are transcribed on four operons in D. circinale AWQC131C with initiation from two bidirectional promoters ( Figure 1). All four transcripts were constitutively expressed in the late exponential growth phase, since sxt mRNA was always detected (data not shown). Transcribed in the reverse direction, operon 1 (sxtDV*EABC; * indicates disrupted ORF of sxtV) spans 7.3 kb and encodes several proteins predicted to be involved early in PST biosynthesis. Operon 2 (sxtPQR) is 3.5 kb and was transcribed in the forward direction.
The catalytic functions of SxtPQR are unknown but are likely to be essential for PST biosynthesis as their presence and organisation is conserved amongst all reported sxt clusters. Operon 3 is monocistronic encoding the predicted permease of the drug/metabolite transporter (DMT) superfamily transporter SxtPER. The largest operon, operon 4, is transcribed in the forward direction and spans 12.8 kb. Operon 4 begins by encoding a protein of unknown function (but conserved in many sxt pathways), Orf24, followed by several proteins involved in PST biosynthesis, resulting in the polycistronic operon orf24sxtSTUNGHMIJKLO.
The 3′ ends of operons 1-4 were bioinformatically screened for putative Rho-dependent and Rhoindependent transcriptional termination sites using the programs TransTerm and TranstermHP, respectively (Kingsford et al., 2007, Jacobs et al., 2009. Rho-independent transcription termination 6 sites were identified in the non-coding region of three out of four mRNA sxt transcripts (Table S1).
Rho-dependent or Rho-independent termination sites could not be identified in the sequence of mRNA encoding operon 1.
Transcription start sites and promoter regions of the sxt operons.
Once the transcription units of the sxt cluster were confirmed, the TSS of each operon was experimentally identified via 5′ RACE (Table 1; Supporting Information). The upstream region of each TSS was screened for a promoter sequence consistent with conserved binding sequences of group 1, 2 and 3 sigma factors (Imamura & Asayama, 2009). All promoters identified in this study displayed sequence similarity to the consensus prokaryote -10 hexamer while there was sporadic presence of the -35 hexamer binding site (Table 1). This indicates that the sxt promoters of D. circinale AWQC131C are activated by an RNA polymerase core enzyme in conjunction with a group 1 or group 2 sigma factor. For the -10 promoter sites identified, a search was conducted for an extended -10 binding site and the upstream (UP) element. The 5′ untranslated region (UTR) region of each operon was also bioinformatically screened for the presence of ribosomal binding sites (RBS), which are periodically present in cyanobacteria (Mutsuda & Sugiura, 2006). Based on the 5′ RACE and bioinformatics data, the sxt gene cluster consists of a total of five TSSs under standard culture conditions ( Figure 2).
Operon 1 begins with the transcription of sxtD and contains a short 5′ UTR of -32 bp in regards to the translation start site and a promoter site showing high similarity to the E. coli σ 70 -10 and -35 hexamers. PsxtP initiates the transcription of operon 2 and was observed to possess a short 5′ UTR spanning 34 bp and containing both -10 and -35 regions. The transcript initiated by PsxtP also displayed a likely RBS with the sequence AAGA six nucleotides upstream of the sxtP translation start site. Expression of the putative transporter sxtPER occurred from two points, PsxtPER and PsxtPER2.
PsxtPER begins -91bp upstream from the annotated translation start site of sxtPER and contains a highly conserved -10 and -35 RNA polymerase binding site. PsxtPER2 is situated +94 bp after the same translational start site. PsxtPER2 contains a highly conserved -10 site, including a single nucleotide seen in extended -10 promoters in addition to a RBS (AAAGAAG). A conserved -35 site was 7 identified 21 bp upstream of the extended -10, showing an unusually long distance between the two hexamers. Porf24, which begins transcription by coding for the hypothetical protein Orf24, has a promoter with a perfectly conserved -10 sequence, including the extended -10 TGn motif. The 5′ UTR for orf24 is 160 bp in length.

Heterologous promoter activity in E. coli
The five promoter regions of the D. circinale sxt cluster identified using 5′ RACE, PsxtP, PsxtD, PsxtPER, PsxtPER2 and Porf24, consisting of 1000 bp from the start codon and downstream, were amplified and cloned into the E. coli vector, pET28b, directly in front of a lux reporter operon replacing the T7 promoter. The plasmids were subsequently transformed into E. coli DH5α. Expression of luciferase from each of these promoters was measured and compared with negative controls; pET28lux harbouring a non-promoter region from within the gene sxtO (pSXL6) and another consisting of the pET28-lux plasmid with no added promoter ( Figure 3). Unpaired t-tests showed that all promoters exhibited significant levels of expression (Table S2) when compared to the negative control (pET28lux).. The heterologous PsxtD, PsxtP, and PsxtPER promoters (pSXL1-3) exhibited the highest levels of luciferase expression in E. coli ( Figure 3A, Table S3). There was a statistically significant difference (p<0.0001) between the highest performing promoter PsxtD (pSXL1) and all the other promoters (pSXL2-5), as well as the controls (pSXL6 and pET28-lux) (Table S4).
When the expression of each promoter was compared to the expression of the non-promoter control sxtO (pSXL6), the strongest promoters (pSXL1-3) had a 1,000 to 10,000-fold increase of luciferase expression ( Figure 3C). A trend was observed in the TSS nucleotide of each operon, with PsxtD, PsxtP, and, PsxtPER having a TSS of adenosine (Table 1). Nucleotide discrimination has been shown to play a role in transcription, with an ATP > GTP > > UTP > CTP initiation sequence preference in both E. coli and Synechocystis PCC6803 (Walker & Osuna, 2002, Mitschke et al., 2011, Kim et al., 2012.
However, the variable transcription of the lux operon cannot be entirely explained by the TSS sequence. The distance of the TSS from -10 region (Pribnow box) has also been shown to be important in transcriptional regulation, with studies using sequence mutations at several positions downstream resulting in various changes in lacZ expression in E. coli (Liu & Turnbough, 1994, Han & 8 Turnbough, 2014.
The expression of PsxtP (pSXL2) is an interesting example of the promoter elements required for heterologous expression of cyanobacterial promoters in E. coli. PsxtP does not seem to have a discernible -35 binding region yet does have a RBS and displayed high expression in E. coli. It is known that the distance between the -10 and -35 effects transcription in cyanobacteria and the -35 hexamer is not always required (Shibato et al., 2002, Subudhi et al., 2008, Dutheil et al., 2012. Thus, the competing preferences between the TSS sequence and position, taken together with other elements of the promoter such as -10, and -35 sequences, transcription factors, the sequence length between the -10 and -35 regions, and the RBS, highlights the complexity of transcriptional regulation and shows the importance of experimental validation of promoter activation data to further improve bioinformatics databases. The promoter responsible for the transcription of orf24 and a second promoter of sxtPER2 (pSXL4 and pSXL5, respectively) expressed lux at levels lower than the other promoters, but still significantly different compared to the controls (Table S4). When compared to sxtO (pSXL6), both promoters showed 12-27-fold increase in expression, as well as 810-and 1770-fold higher expression than the pET28-lux control, respectively ( Figure 3B; Table S3). These results indicated that the promoters appear to be active, though at lower levels than the other three promoters ( Figure 3B).
Both bidirectional promoters of the D. circinale AWQC131C sxt pathway were active and able to drive the expression of the luciferase operon in E. coli. However, while the three promoters PsxtD, PsxtP and PsxtPER are expressed in E. coli, low levels of expression from Porf24 could be a potential bottleneck in the heterologous expression of operon 4 which encodes the majority of genes required for PST biosynthesis, making it a prime target for promoter switching for optimising STX production.
The presence of two TSSs, PsxtPER and PsxtPER2, located up-and down-stream of the annotated translation start sites is unusual and may result in two forms of SxtPER. While PsxtPER2 was able to induce expression of the luciferase operon, it was at significantly lower levels than PsxtPER. A putative NtcA-binding site was found to overlap the -35 position of the PsxtPER2 promoter, and may have affected in vivo activation of the promoter in E. coli. Therefore, we cannot conclusively 9 determine if this is in fact a real promoter and two isoforms of SxtPER are translated. Interestingly, BLASTp analysis revealed that a complete RhaT super domain of the drug/metabolite transporter family is present within the shorter isoform. In the event that sxtPER is translated as two isoforms, it is unclear what functionality this would provide, but may alter substrate specificity and enable the export of different PST analogues, which may additionally be regulated by NtcA binding. Further experimental work in the host, D. circinale will be required to determine the role of the gene regulated by PsxtPER2.

The promoters that regulate saxitoxin biosynthesis, showed lowered expression levels in
Synechocystis PCC6803 compared to E. coli. The surprising nature of this result has previously been observed by the zinc inducible promoter Psmt from Synechococcus sp. PCC7002. Psmt resulted in stronger induction of protein synthesis for ethylene production in E. coli compared to the residual levels observed in Synechocystis PCC6803 (Guerrero et al., 2012). It is possible that the higher level of transcription inhibits biosynthesis in the heterologous host. For example, heterologous expression of lyngbyatoxin (ltxA-D) in E. coli was only successful under the weaker Ptet O promoter compared to a T7 promoter (Ongley et al., 2013). The low level expression by the Ptet O promoter system has since been successful for the heterologous expression of multiple cyanobacterial biosynthetic gene clusters in E. coli (Liu et al., 2017, Greunke et al., 2018. The sxtD promoter is responsible for the transcription of operon 1 of the D. circinale sxt cluster, which carries the core biosynthetic genes including the polyketide-like enzyme sxtA. PsxtD (SXL7) showed very low expression levels that were only 1.3-fold higher than the control strain. The lack of statistically significant expression of PsxtD (SXL7) has adverse implications for heterologous expression of PSTs using Synechocystis PCC6803 as a heterologous host. Similar to Porf24 in E. coli, PsxtD is a candidate for promoter exchange for optimised heterologous expression.
PsxtP (SXL8) and Porf 24 (SXL10) showed consistent level of luciferase expression per OD 730 throughout the experiment ( Figure 4C). The promoter responsible for permease expression (PsxtPER) had initial levels that were up to three-fold higher than PsxtP (SXL8) but decreased over the course of growth ( Figure 4C). This is consistent with previous studies that indicated both intracellular and extracellular saxitoxin levels in D. circinalis AWQC131C were consistent throughout exponential growth phase (Ongley et al., 2016). Higher expression of SxtPER during early stages of growth prevents the internal toxin accumulation leading to the consistent levels of intracellular and extracellular saxitoxin.

Conclusion
PSTs have a range of biotechnology applications and manipulation of the gene cluster for expression in a heterologous host will improve availability of compounds for these uses. This study experimentally revealed D. circinale AWQC131C sxt promoters and tested their activity in E. coli and Synechocystis PCC6803 hosts. In E. coli, the PsxtD, PsxtP, PsxtPER promoters exhibited higher levels of luciferase expression while the Porf24 promoter expression was at a lower level. Therefore, future design a D. circinale AWQC131C sxt gene cluster construct to be cloned directly into E. coli with the Porf24 promoter being the target of promoter exchange with an alternative promoter. Similarly, PsxtD is the candidate for promoter exchange in Synechocystis PCC6803 in future cloning experiments. In E.
coli, the stx promoters appeared to result in a greater transcriptional increase compared with the control, however, further analysis is required to determine if this organism will be the better 11 heterologous host.
Promoter characterisation and analysis of the D. circinale AWQC 131C sxt gene cluster has identified heterologous host dependent bottlenecks in PST biosynthesis. This study demonstrates the potential use of native promoters for the heterologous expression of STX and other pharmaceutically important cyanobacterial metabolites.

Methods
Strains and culture conditions D. circinale AWQC131C was maintained in Jaworski's Medium (JM) (Thompson et al., 1988) at 24°C ±1°C and illuminated with 11 μmol m -2 s -1 of photons on a 12:12 h light/dark cycle. Unless otherwise specified, E. coli strains (Table S8)were maintained in Luria broth or on agar plates supplemented with 100 µg mL -1 ampicillin or 50 µg mL -1 kanamycin and grown at 37°C. Synechocystis sp. PCC6803 was maintained in BG-11 media supplemented with 100 µg mL -1 spectinomycin when required, at 30°C under constant illumination.

Total RNA extraction, cDNA synthesis and transcriptional analysis
To extract high quality total RNA, cell pellets were snap frozen in liquid nitrogen and ground into a fine powder with a mortar and pestle prior to extraction with the RNeasy Plant Mini kit (QIAGEN).
Residual gDNA was removed from total RNA samples using TURBO DNA-free™ DNase as described by the manufacturer (Ambion). Removal of contaminating gDNA was validated via PCR with the 27F/809R PCR primer set targeting the cyanobacterial 16S rRNA gene (Jungblut et al., 2005). RNA quality was also checked by formaldehyde gel electrophoresis, while gDNA was checked by agarose gel electrophoresis.
The Superscript ® III First Strand synthesis system (Invitrogen) was used to reverse transcribe 1 µg of total RNA primed with an antisense gene specific primer (GSP). Transcriptional units were determined by PCR amplification in a 20 μL reaction mixture containing 2.5 mM MgCl2, 1 × PCR buffer (Fisher Biotec, Geneworks), 10 pmol dNTPs (Austral Scientific), 10 pmol of GSP, 1 U of Taq polymerase (Fisher Biotec, Geneworks) and sterile Milli-Q water. Thermal cycling was performed in a Bio-Rad 96-well iCycler (Bio-Rad) and began with an initial denaturation cycle of 95°C for 4 min, followed by 35 cycles of DNA denaturation at 95°C for 20 s and primer annealing at 55°C for 20 s. DNA strand extension was altered to 1 min for every 1 kb of amplified product. A final extension at 72°C for 7 min and a final holding temperature of 4°C completed the thermal cycling. Each reaction contained cDNA as the template and two primers (Table S9), each designed to target an adjacent gene. Amplification was observed if the two adjacent genes were located on the same mRNA transcript. A positive control for each PCR contained gDNA. Two negative control reactions were performed by adding template from a cDNA synthesis reaction, the first omitting reverse transcriptase and the second reaction omitting a nucleic acid template.
Isolation D. circinale AWQC131C sxt gene cluster transcription start sites and promoters using 5′RACE To isolate the promoter of each transcriptional unit, transcription start sites (TSSs) were localised with the FirstChoice ® RLM-RACE kit for 5′ rapid amplification of cDNA ends (5′ RACE) (Ambion) with 10 µg total RNA as starting material. The 5′ RACE adapter was ligated directly onto RNA followed by reverse transcription cDNA synthesis. First round PCR reactions were performed using a 5′ outer adapter primer in conjunction with four reverse GSPs at approximately 50-100 bp intervals ( Figure S1, Table   S10). Reactions containing amplified products from the first round PCR became the template for second round nested PCR a 5′ adapter inner primer in conjunction with the same four reverse primers.
Amplicons of interest were analysed on a 2% (w/v) agarose gel and purified using a QIAquick spin gel extraction kit (QIAGEN). Purified PCR products were then cloned into the pGEM-T Easy vector (Promega) and sequenced using an ABI 3730 capillary sequencer at the Ramaciotti Centre DNA Sequencing Analysis Facility, UNSW.

Cloning and transformation
The TOPO TA cloning ® kit (Invitrogen) and the pGEM ® -T Easy Vector kit (Promega) were used for the cloning and transformation of E. coli (Table S8). Cloning with the TOPO TA cloning ® kit involved setting up a ligation reaction containing 4 µL of PCR product, 1 µL of Invitrogen salt solution (1.2 M NaCl, 0.06 M MgCl 2 ) and 10 ng pCR ® 2.1-TOPO ® plasmid DNA (Invitrogen). The ligation reaction was incubated for 20 min at room temperature and was then ready for transformation. The pGEM ®-T Easy 13 vector ligation reaction contained 1 × rapid ligation buffer (Promega), 50 ng of pGEM ®-T Easy vector DNA (Promega), 3 Weiss U of T4 DNA ligase (Promega) and 3 µL of PCR product. The ligation reaction was left to incubate overnight at 4°C and was then ready for transformation. Colonies were selected by blue and white screening and the presence of an insert was confirmed by colony PCR using either the primer sets M13F and M13R (pCR ® 2.1-TOPO) or T7F and M13R (pGEM-T Easy). Plasmids containing an insert were then sequenced.
Promoter expression in E. coli Five promoters (PsxtD, PsxtP, PsxtPER, Porf24, PsxtPER2) and a non-promoter region within the sxtO open reading frame were cloned into the pET28b expression vector along with the luciferase reporter operon from Photorhabdus luminescens ( Figure S4A). The luciferase operon (luxCDABE) was amplified from the pLuxNSII plasmid (Woelfle et al., 2007) (denaturation at 98°C for 3 min, followed by thirty cycles of denaturation at 98°C for 15 s, annealing step at 60°C for 20s, extension at 72°C for 30s/kb, and a final extension at 72°C for 10 min), and the pET28b backbone was also PCR amplified to remove the T7 promoter region. All primers were designed using the NEBuilder assembly tool (Table   S11). Double-stranded PCR fragments were amplified using the KAPA HiFi Hotstart DNA polymerase (KAPA biosystems). The pET28b backbone, sxt promoter (Psxt), and luciferase (lux)fragments were assembled using the Gibson assembly master mix (NEB) (Gibson et al., 2009), and incubated at 50°C, for 1 h. The reaction was transformed into chemically competent E. coli DH5α and positive colonies selected.

Promoter expression in Synechocystis PCC6803
The Psxt-lux integration vector was constructed through classical restriction/ligation cloning using the restriction enzymes NotI and KpnI (NEB). The Psxt-lux fragments were amplified from the pET28b-Plux vectors using the lacI-P-lux_NotI_F, and lacI-P-lux_KpnI_R primers (Table S11). Linear DNA fragments were digested, purified and ligated into the pSYN_6.3 vector ( Figure S4B) using T4 DNA ligase at 22°C for 1 h, followed by transformation into E. coli DH5α and colony screening. Plasmid constructs were confirmed by terminal-end sequencing.
The integration of the Psxt-lux fragments into the Synechocystis PCC6803 genome ( Figure S4C) was 14 achieved via the natural competence of Synechocystis PCC6803 (Eaton-Rye, 2011). Synechocystis PCC6803 was grown at 30°C, shaking at 100 rpm, under constant light until exponential phase, and used to inoculate a 50 mL flask to an initial OD 730 of 0.05. After ~4 days of photoautotrophic growth, the cells reached an OD 730 of 0.5, and cells harvested in a by centrifugation at 2750 g for 5 min. Cells were resuspended in 2 mL of fresh BG11 medium, divided into 0.5 mL aliquots (OD 730 of 2.5), combined with 10 µg of DNA and incubated at 30°C for 6 h. A sterile Immobilon Transfer membrane (Merk Millipore) was placed on each BG11 agar plate, spread with 200 µL of the transformation mixture and incubated for 12 h under light at 30°C. Membranes were transferred to BG11 agar plates containing 25 µg mL -1 spectinomycin. The plates were incubated for a further two days at 30°C under constant illumination, before the membrane was transferred to plates containing 50 µg mL -1 spectinomycin and incubated for a further 7-10 days until colonies become visible. Recombinant Synechocystis 6803 colonies were picked and streaked on BG-11 agar plates supplemented with 100 µgmL-1 spectinomycin for three rounds to achieve full segregation of the integration. Transformants were confirmed using the PhaCaF and PhaCbR PCR primers (Table S10).

Psxt activity in E. coli DH5α
Luciferase constructs were transformed into E. coli DH5α and grown on M9 minimal medium supplemented with 50 μg mL -1 Kanamycin, at 37˚C for 24 h. Luminescence and optical density measurements were taken at one-hour intervals until the OD 600 reached 0.8. A final measurement was taken at 24 h (PsxtD n = 6, PsxtP n = 6, PsxtPER n = 6, Porf24 n = 3, PsxtPER2 n = 6, sxtO n = 6, pET28-lux n = 9). The strength of each promoter was measured as the highest rate of luminescence change, normalised to a change in OD 600 . One-way ANOVA (Graphpad Prism 7) was used to calculate any statistical differences between the promoters. Unpaired t-tests were also used to determine the statistical differences between the strains and the control.

Psxt activity in Synechocystis PCC6803
Transformed Synechocystis PCC6803 strains were inoculated at an OD 730 of 0.05 and grown in BG-11 media with the 100 µg mL -1 spectinomycin at 30°C, shaking in constant light. OD 730 and relative light 15 units were measured every 24 h for 400 h (PsxtD n = 3, PsxtP n = 3, PsxtPER n = 3, Porf24 n = 3, Syn6803-lux n = 3). Promoter strength was measured by determining the highest luminescence per OD 730 and statistically assessed for expression in Synechocystis PCC6803. One-way ANOVA (Graphpad Prism 7) was used to calculate any statistical differences between the promoters. Unpaired t-tests were also used to determine the statistical differences between the strains and the control.

Declarations
Ethics approval and consent to participate Not applicable Consent for publication Not applicable Availability of data and material All data generated or analysed during this study are included in this published article and its supplementary information files.

Competing interests
The authors declare that they have no competing interests